[Shootout-list] Numerical medium-sized benchmarks

[Brent Fulgham]

--- Sebastien Loisel <sloisel@gmail.com> wrote:
> I've noticed that a (large?) fraction of game 
> software has needs that are similar to scientific 
> computing. For instance, the physics simulation
> is basically an ODE solver, and the 3d math usually
> involves very long vectors of floating point
numbers.
> So even though my good friend Patrick 
> (http://plam.csail.mit.edu/) assures me that
> my requirements are niche at best, I am convinced 
> that the game people (which some argue drive 
> performance pc computing) have similar
> requirements.

I agree.  Game physics requires tremendous processing
power, and is (aside from the new home multimedia 
authoring push) one of the major drivers of the
current processor evolution.  (We don't need much more
power than we have today to drive web browsers and
view word documents!)

> Unfortunately, the performance was disappointing
> (see for instance
http://groups-beta.google.com/group/comp.lang.functional/browse_frm/thread/4817bd07ade7a9f9/ebc182b6cd6005c8).

This is a dead link for me (shows up as Google not
knowing what this topic is).  Can you provide a
textual search refernce for Google?  When I viewed
comp.lang.functional it provided a different URL
format, so perhaps they have changed something on
their back-end recently.

> The ML code ran in more than twice the time the C++
> code took, and Haskell was about eleven times
slower,
> and the code I had wasn't yet at the level of 
> abstraction I needed. Furthermore, it started
looking
> as if the more abstract code I intended to write
> might be harder to write in ML and Haskell.

This is a great finding.  Most of the languages
in the shootout do reasonably well in the various
problems on the site.  Perhaps its time to push
into new areas so that performance can be examined
in other problem domains.

-Brent